Module transcript 

Transcript linking and token extraction.

Reads Claude Code transcript files (JSONL), extracts token usage (deduplicated by message.id), modified files, and spawned agent IDs.

Structs

ConversationMessage

A conversation message suitable for display.

DiffResult

ThinkingBlock

A captured thinking/reasoning block from an assistant message.

TranscriptAnalysis

Results from parsing a transcript.

Functions

archive_transcript_to_git

Copy a transcript file into .git/chub/transcripts/ for local review. The copy is stored as <session_id>.jsonl inside the repo’s .git directory so it won’t be committed but is available for LLM review.

calculate_attribution

Calculate line attribution between two commits.

claude_projects_dir

Find the Claude Code projects directory for a given repo path. Returns ~/.claude/projects/<sanitized-repo>/.

find_transcript

Find a Claude Code transcript file for a session.

get_diff_files

Get files modified since a given commit using git diff-tree.

parse_conversation

Parse a transcript into displayable conversation messages. Filters out system messages and deduplicates by UUID.

parse_transcript

Parse a Claude Code transcript file and extract token usage and metadata. Deduplicates by message.id (Claude streams may create multiple entries).

parse_transcript_range

Parse a range of a transcript (from byte offset, optionally limited lines).

transcript_line_count

Count lines in a transcript file.

transcript_size

Get the byte size of a transcript file.